Limits of detection

All limits were calculated based on the level of analytes in the instrumental blanks, which were described section 3.6. The calculated result are presented in Table 4.5, and the raw data are presented in Table B.3 in Appendix B. The instrumental limit of detection (IDL) ranged from 0.05-0.79 pg for PFCs and 1.22-22.8 pg for VOCs. The method LOQ ranged from 0.06-0.69 ng/m³for PFCs and 1.81-16.1 ng/m³for VOC.

For conditioning of tubes EPA (1999b) recommends prepacked tubes to be condi-tioned for 30 minutes at 350°C while at least 50 mL/min carrier gas flow through. But for this study, each tube was conditioned at 300 °C over 30 minutes before used as blanks or sample tubes. The temperature was lowered to 300 °C because of the maximum temper-ature for Tenax TA, 350°C (Dettmer and Engewald, 2002). This should be sufficient for conditioning of the tubes and minimize the contamination issues.

Table 4.5: Limit of detection and quantification

The VOCs had all a higher level of blank contamination and this could be explained by that these compounds exist around because of origin of emissions.

4.4.4 Recovery

Apparent recoveries for 500 pg, 200 pg, and the absolute recovery for the desiccator-test at 500 pg are presented in Table 4.6. Numbers for apparent recovery are corrected with the response of the ISTD, while absolute recoveries are not corrected with the response of the ISTD, calculated as described in the Section 3.2.

Apparent recoveries were also tested at 10 pg. There results were rejected because the recoveries ranged from 311.3 to 1099.5%, numbers presented in B.4. For the compounds α -pinene and β -pinene, 10 pg are below the IQL. At low concentrations the reliability of the results is low. Skoog et al. (2014) explains that “at ultratrace levels of 1 ppb, interlaboratory error (%RSD) is nearly 50%. At lower levels, the error approaches 100%.”

This could be the explanation of the very high recoveries at 10 pg.

One replicate was an obvious outlier for both 500 and 200 pg, and where because of that ignored when recoveries were determined. This resulted in apparent recoveries at 500 and 200 pg only have two replicates. The raw data for all the replicates is shown in Table B.4, where replicate named mix-200 and mix3-500 were excluded. This was because the signal from the ISTD was high compared to the ISTD signal in the other replicates resulting in minimal concentrations for the analytes, as these areas held the same order of magnitude as in the other replicates. This could be caused by a volume error during the spiking of these tubes, since they were manually spiked with Hamilton syringes.

Hexanal was excluded from sample results due to recoveries for 500 pg at 30.0 ± 141% and for 200 pg, 240 ± 4%. There is no other compound that deviates this much in the 500, 200 pg replicates. As explained earlier, biogenic VOCs i are compounds that are around us at any time. The unaccepted recoveries for hexanal could be attributed to contamination issues as this compound most probably occurs in the lab atmosphere.

Jiang et al. (2017) found that hexanal were the major odorous compounds emitted from

particleboard, which are a common building material in the lab as well. Further, hexanal had, according to the calibration curves, the best sensitivity of all compounds. This can contribute to higher sensitivity for low contamination levels.

Table 4.6: Apparent recovery and RSD [%] for 500 pg and 200 pg, and absolute recovery and RSD [%] at 500 pg spiked in a desiccator

Recovery ± RSD Recovery ± RSD Recovery-desiccator ± RSD Analyte [%] 500 pg n=2 [%] 200 pg n=2 [%] 500 pg n=3

At 500 pg, all the other compounds showed acceptable recoveries. N-EtFHxSE had an average apparent recovery at 40.4 ± 22.7%. This is on the edge of the validation limits set for this method. However, the result is an average of just two replication,

α -pinene, β -pinene and β -myrcene all had apparent recoveries above the acceptable limit at 200 pg. Again, this could be caused by contamination in the lab or the instru-ment room. These levels of contamination are quite challenging to control because of the natural emissions of the VOCs.

The desiccator experiment was conducted with spiking a desiccator at ambient tem-perature, waiting at least 30 minutes, and then withdrawal of the air with the sampling pump. It was not possible to connect this system to clean nitrogen gas or another clean gas. This caused the pump to stop after few seconds because of under-pressure in the system. The valve, closing the desiccator, needed to be slightly open so that the pressure was constant in the desiccator. This causes potential errors. First of all, there is the poten-tial that the volatile analytes evaporates through the slightly open valve, because of their vapor pressure.

Toluene-D8 have a high vapor pressure, evaporation out of the desiccator could cause the areas to be much lower than for the tubes spiked with liquid at the same concentration.

Because of this, the recoveries were calculated as absolute recovery, without correcting them for the response of the ISTD. The absolute recovery for all compounds, except for FMBrBz and N-EtFHxSE, is much higher than the accepted range. This indicates that the air in the lab environment actually are contaminated with biogenic compounds as suggested earlier.

4.4.5 Breakthrough and carry-over

In the spiked desiccator, the breakthrough was tested by having two tubes in tandem when sampling.This was done in three replications. The results are presented in figure 4.4.

In the desiccator test, all tubes were analyzed beforehand to be sure that target analyte levels were low to zero. The raw results are included in B.2 in Appendix B. There were reported levels in almost all tubes, as for other lab blanks. These levels were not corrected in the breakthrough test, and therefore, could contribute to higher breakthrough.

The calculated breakthrough showed that only BrFBz was below 5 % with 0.0 ± 0.0

%, followed by Hexanal with 6.64 ± 34.22 %. The remaining analytes varied from 10.37

± 17.27 % to 50.36 ± 173.21 %. As described for the recoveries for the desiccator, the set-up of the sampling could cause contamination of the samples because air were let in the desiccator. Since the SSV was not calculated and the order of the potential contamination is unknown, it is not possible to know if the sorbent media in the first tube was saturated, causing the analyte molecules to be collected in the second tube. This could explain the high breakthrough for the monoterpenes.

N-EtFHxSE has not only a considerable high breakthrough, but also a very high RSD.

Compared to the absolute recovery for N-EtFHxSE, which only were at 15.6 %, with a RSD at 173.2%. There is a high variation in both tubes in tandem, within all three replicates. These variations can be explained that N-EtFHxSE were detected in all sample tubes, but had low response, resulting a calculated concentration of zero for two of the replicates. This result indicates that the sampling of N-EtFHxSE war not compleeted, and that N-EtFHxSE either evaporized out of the chamber or did not vaporize in the chamber.

These variations could also be caused by contamination. However, N-EtFHxSE is not a compound that have been frequently detected in indoor air, which make a contamination issue less plausible.

When checking the carry-over, the calibration tubes were re-analyzed. In Table 4.7, the results are presented.

At higher concentrations there is no carry-over. But for the lower concentrations, es-pecially the 1 pg, there is some carry-over. The results marked with superscript 1 had concentrations in the second round of analysis exceed the concentration in the first anal-ysis. As explained earlier, this could be caused by contamination of the tubes. When conducting these analyses, all tubes were loaded into the carousel of the ATD. One by one, the tubes were analyzed and then loaded back into the carousel while the rest of the analysis was run. Between 12 to 16 tubes were analyzed in one run, resulting in an analy-sis time around 12-14 hours before the analyanaly-sis was started again to check for carry-over in the tubes. This causes the tubes to be stored in an environment with no control over possible contamination. As explained earlier, at low concentrations, a small change in response can cause then give significant concentration changes.

When excluding the internal standard, new calibration curves needed to be made for all analytes. For many compounds, these were less linear. These calibration curves are

Figure 4.4: Breakthrough [%] in the desiccator-test. The error-bars represents the RSD [%]. The max line are at 5%.

Table 4.7: The ratio [%] between first and second adsorption for the calibration curve standards

1 pg 10 pg 50 pg 100 pg 200 pg 400 pg 800 pg 1000 pg

BrFBz 1.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0

FMBrBz 0.5 0.18 0.0 0.0 0.0 0.0 0.0 0.0

α -Pinene 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

β -Pinene 1423.5¹ 0.0 0.0 0.0 0.0 0.0 0.0 0.0

β -Myrcene 2688.5¹ 0.0 0.0 0.0 0.0 0.0 0.0 0.0

N-EtFHxSE 0¹ 0.0 0.0 0.0 0.0 0.0 0.0 0.0

listed in Figure C.2 in Appendix C. This will then make the test results without correction of the response of the ISTD less reliable. However, it will still indicate whether there was breakthrough or not.

The carry-over test showed that the ATD successfully desorbed the analytes retained in each tube. There were some inconclusive results for 1 pg, but since this was the lowest calibration level, its also the level with the highest uncertainties (Skoog et al., 2014).